def createHistograms(self, hist_cfg, all_stack=False, verbose=False, friend_func=None, vcfgs=None):
'''Method to create actual histogram (DataMCPlot) instances from histogram
config; this version handles multiple variables via MultiDraw.
'''
# set_trace()
pool = Pool(processes=len(self.hist_cfg.cfgs))
print('number of processes for filling histos (ie. samples): %i'%len(self.hist_cfg.cfgs))
result = pool.map(self.makealltheplots, self.hist_cfg.cfgs) # for no return
# result = pool.apply_async(self.makealltheplots, self.hist_cfg.cfgs) # for no return
# self.plots = result.get()
# set_trace()
for i, cfg in enumerate(self.hist_cfg.cfgs):
stack = not cfg.is_data and not cfg.is_signal
# print(cfg.name, stack)
for vcfg in self.vcfgs:
hist = result[i][vcfg.name].histos[0].obj # result[0]['CR_hnl_m_12'].histos[0]
# hist = hists[vcfg.name]
plot = self.plots[vcfg.name]
# hist.Scale(cfg.scale)
# set_trace()
if cfg.name in plot:
print 'Histogram', cfg.name, 'already exists; adding...', cfg.dir_name
hist_to_add = Histogram(cfg.name, hist)
if not cfg.is_data:
hist_to_add.SetWeight(hist_cfg.lumi*cfg.xsec/cfg.sumweights)
plot[cfg.name].Add(hist_to_add)
else:
# print(cfg.name, hist.GetEntries(), stack)
plot_hist = plot.AddHistogram(cfg.name, hist, stack=stack)
# print('added histo %s'%vcfg.name)
if not cfg.is_data:
plot_hist.SetWeight(self.hist_cfg.lumi*cfg.xsec/cfg.sumweights)
# print(cfg.name, vcfg.name, len(plot.histos))
print('initializing histos done, making stacks...')
for plot in self.plots.itervalues():
plot._ApplyPrefs()
print('number of processes for drawing (ie. stacks to draw): %i'%len(self.plots))
procs = []
for i, plot in enumerate(self.plots.itervalues()):
proc = Process(target=plot.Draw, args=())
procs.append(proc)
proc.start()
for proc in procs:
proc.join()
# for plot in self.plots.itervalues():
# plot._ApplyPrefs()
# plot.Draw()
return self.plots
def AddHistogram(self, name, histo, layer=0, legendLine=None):
'''Add a ROOT histogram, with a given name.
Histograms will be drawn by increasing layer.'''
tmp = Histogram(name, histo, layer, legendLine)
self.histos.append(tmp)
self.histosDict[name] = tmp
def makealltheplots(self, cfg):
verbose=False
friend_func=None
all_stack=False
# for cfg in [hist_cfg.cfgs[0]]:
# for cfg in hist_cfg.cfgs:
# First check whether it's a sub-histo or not
if isinstance(cfg, HistogramCfg):
hists = createHistograms(cfg, all_stack=True, vcfgs=self.vcfgs)
for h in hists: print(h)
for vcfg in self.vcfgs:
hist = hists[vcfg.name]
plot = self.plots[vcfg.name]
set_trace()
hist._BuildStack(hist._SortedHistograms(), ytitle='Events')
print('stack built')
total_hist = plot.AddHistogram(cfg.name, hist.stack.totalHist.weighted, stack=True)
if cfg.norm_cfg is not None:
norm_hist = createHistogram(cfg.norm_cfg, all_stack=True)
norm_hist._BuildStack(norm_hist._SortedHistograms(), ytitle='Events')
total_hist.Scale(hist.stack.integral/total_hist.Yield())
if cfg.total_scale is not None:
total_hist.Scale(cfg.total_scale)
# print 'Scaling total', hist_cfg.name, 'by', cfg.total_scale
else:
# print('building histgrams for %s'%cfg.name)
# It's a sample cfg
# Now read the tree
if cfg.dir_name == "DDE":
# file_name = '/'.join([cfg.ana_dir, cfg.dir_name, cfg.tree_prod_name, 'tree_fr_DR_data_v2.root'])
file_name = '/'.join([cfg.ana_dir, cfg.dir_name, cfg.tree_prod_name, 'tree_fr_DR_data_v2_oldVars.root'])
else:
file_name = '/'.join([cfg.ana_dir, cfg.dir_name, cfg.tree_prod_name, 'tree.root'])
# attach the trees to the first DataMCPlot
plot = self.plots[self.vcfgs[0].name]
try:
ttree = plot.readTree(file_name, cfg.tree_name, verbose=verbose, friend_func=friend_func)
except:
set_trace()
norm_cut = self.hist_cfg.cut
shape_cut = self.hist_cfg.cut
if cfg.norm_cut:
# norm_cut = cfg.norm_cut
norm_cut += cfg.norm_cut # to add met filters only for data
# print('sample = %s, cuts = %s'%(cfg.name, norm_cut))
if cfg.shape_cut:
shape_cut = cfg.shape_cut
weight = self.hist_cfg.weight
if cfg.weight_expr:
weight = '*'.join([weight, cfg.weight_expr])
if cfg.cut_replace_func:
norm_cut = cfg.cut_replace_func(norm_cut)
shape_cut = cfg.cut_replace_func(norm_cut)
#if needed, apply the loose selection for the non-prompt region
if cfg.is_dde == True:
norm_cut = norm_cut.replace('l1_reliso_rho_04 < 0.15 & l2_reliso_rho_04 < 0.15 & hnl_iso04_rel_rhoArea < 1','(l1_reliso05 > 0.15 | l2_reliso05 > 0.15) & hnl_iso04_rel_rhoArea < 1')
shape_cut = shape_cut.replace('l1_reliso_rho_04 < 0.15 & l2_reliso_rho_04 < 0.15 & hnl_iso04_rel_rhoArea < 1','(l1_reliso05 > 0.15 | l2_reliso05 > 0.15) & hnl_iso04_rel_rhoArea < 1')
if 'Conversion' in cfg.name:
norm_cut = norm_cut + ' & (l0_gen_match_pdgid == 22 | l1_gen_match_pdgid == 22 | l2_gen_match_pdgid ==22)'
shape_cut = shape_cut + ' & (l0_gen_match_pdgid == 22 | l1_gen_match_pdgid == 22 | l2_gen_match_pdgid ==22)'
if self.hist_cfg.weight:
norm_cut = '({c}) * {we}'.format(c=norm_cut, we=weight)
shape_cut = '({c}) * {we}'.format(c=shape_cut, we=weight)
#insert the fake rate weight for doublefakes
if cfg.is_dde == True and cfg.is_doublefake == True:
norm_cut = '({c}) * {we}'.format(c=norm_cut, we='weight_fr/(1-weight_fr)')
shape_cut = '({c}) * {we}'.format(c=shape_cut, we='weight_fr/(1-weight_fr)')
#insert the fake rate weight for singlefakes
if cfg.is_dde == True and cfg.is_singlefake == True:
norm_cut = '({c}) * {we}'.format(c=norm_cut, we='((weight_fr/(1-weight_fr))-((weight_fr/(1-weight_fr))*(weight_fr/(1-weight_fr))))')
shape_cut = '({c}) * {we}'.format(c=shape_cut, we='((weight_fr/(1-weight_fr))-((weight_fr/(1-weight_fr))*(weight_fr/(1-weight_fr))))')
# print '#### FULL CUT ####', norm_cut
#adapt branch names to different software versions
if not ttree.FindBranch("l0_reliso_rho_04"):
norm_cut = norm_cut.replace('l0_reliso_rho_04','l0_reliso05')
shape_cut = shape_cut.replace('l0_reliso_rho_04','l0_reliso05')
norm_cut = norm_cut.replace('l1_reliso_rho_04','l1_reliso05')
shape_cut = shape_cut.replace('l1_reliso_rho_04','l1_reliso05')
norm_cut = norm_cut.replace('l2_reliso_rho_04','l2_reliso05')
shape_cut = shape_cut.replace('l2_reliso_rho_04','l2_reliso05')
if not ttree.FindBranch("hnl_iso04_rel_rhoArea"):
norm_cut = norm_cut.replace('hnl_iso04_rel_rhoArea','hnl_iso_rel')
shape_cut = shape_cut.replace('hnl_iso04_rel_rhoArea','hnl_iso_rel')
# Initialise all hists before the multidraw
hists = {}
for vcfg in self.vcfgs:
hname = '_'.join([self.hist_cfg.name, hashlib.md5(self.hist_cfg.cut).hexdigest(), cfg.name, vcfg.name, cfg.dir_name])
if any(str(b) == 'xmin' for b in vcfg.binning):
hist = TH1F(hname, '', vcfg.binning['nbinsx'],
vcfg.binning['xmin'], vcfg.binning['xmax'])
else:
hist = TH1F(hname, '', len(vcfg.binning)-1, vcfg.binning)
initHist(hist, vcfg)
hists[vcfg.name] = hist
var_hist_tuples = []
for vcfg in self.vcfgs:
var_hist_tuples.append('{var} >> {hist}'.format(var=vcfg.drawname, hist=hists[vcfg.name].GetName()))
# Implement the multidraw.
try:
ttree.MultiDraw(var_hist_tuples, norm_cut)
except:
set_trace()
# Do another multidraw here, if needed, and reset the scales in a separate loop
if shape_cut != norm_cut:
scale = hist.Integral()
ttree.Project(hname, vcfg.drawname, shape_cut)
try:
hist.Scale(scale/hist.Integral())
except:
set_trace()
stack = all_stack or (not cfg.is_data and not cfg.is_signal)
# set_trace()
# Loop again over the variables and add histograms to self.plots one by one
for vcfg in self.vcfgs:
hist = hists[vcfg.name]
plot = self.plots[vcfg.name]
hist.Scale(cfg.scale)
if cfg.name in plot:
# print 'Histogram', cfg.name, 'already exists; adding...', cfg.dir_name
hist_to_add = Histogram(cfg.name, hist)
if (not cfg.is_data) and (not cfg.is_dde):
hist_to_add.SetWeight(self.hist_cfg.lumi*cfg.xsec/cfg.sumweights)
if cfg.is_dde:
# hist_to_add.SetWeight(hist_to_add.obj.GetEntries()/cfg.sumweights)
hist_to_add.SetWeight(cfg.sumweights)
plot[cfg.name].Add(hist_to_add)
else:
# print(cfg.name, hist.GetEntries(), stack)
plot_hist = plot.AddHistogram(cfg.name, hist, stack=stack)
# print('added histo %s for %s'%(vcfg.name,cfg.name))
if (not cfg.is_data) and (not cfg.is_dde):
plot_hist.SetWeight(self.hist_cfg.lumi*cfg.xsec/cfg.sumweights)
if cfg.is_dde:
try:
# plot_hist.SetWeight(plot_hist.obj.GetEntries()/cfg.sumweights)
plot_hist.SetWeight(cfg.sumweights)
except:
set_trace()
print('added histograms for %s'%cfg.name)
PLOTS = self.plots
return PLOTS
def createHistogram(hist_cfg, all_stack=False, verbose=False, friend_func=None):
'''Method to create actual histogram (DataMCPlot) instance from histogram
config.
'''
plot = DataMCPlot(hist_cfg.var.name)
plot.lumi = hist_cfg.lumi
vcfg = hist_cfg.var
for cfg in hist_cfg.cfgs:
# First check whether it's a sub-histo or not
if isinstance(cfg, HistogramCfg):
hist = createHistogram(cfg, all_stack=True)
hist._BuildStack(hist._SortedHistograms(), ytitle='Events')
total_hist = plot.AddHistogram(cfg.name, hist.stack.totalHist.weighted, stack=True)
if cfg.norm_cfg is not None:
norm_hist = createHistogram(cfg.norm_cfg, all_stack=True)
norm_hist._BuildStack(norm_hist._SortedHistograms(), ytitle='Events')
total_hist.Scale(hist.stack.integral/total_hist.Yield())
if cfg.total_scale is not None:
total_hist.Scale(cfg.total_scale)
else:
# It's a sample cfg
hname = '_'.join([hist_cfg.name, hashlib.md5(hist_cfg.cut).hexdigest(), cfg.name, vcfg.name, cfg.dir_name])
if any(str(b) == 'xmin' for b in vcfg.binning):
hist = TH1F(hname, '', vcfg.binning['nbinsx'],
vcfg.binning['xmin'], vcfg.binning['xmax'])
else:
hist = TH1F(hname, '', len(vcfg.binning)-1, vcfg.binning)
initHist(hist, vcfg)
file_name = '/'.join([cfg.ana_dir, cfg.dir_name, cfg.tree_prod_name, 'tree.root'])
ttree = plot.readTree(file_name, cfg.tree_name, verbose=verbose, friend_func=friend_func)
norm_cut = hist_cfg.cut
shape_cut = hist_cfg.cut
if cfg.norm_cut:
norm_cut = cfg.norm_cut
if cfg.shape_cut:
shape_cut = cfg.shape_cut
if cfg.cut_replace_func:
norm_cut = cfg.cut_replace_func(norm_cut)
shape_cut = cfg.cut_replace_func(norm_cut)
weight = hist_cfg.weight
if cfg.weight_expr:
weight = '*'.join([weight, cfg.weight_expr])
if hist_cfg.weight:
norm_cut = '({c}) * {we}'.format(c=norm_cut, we=weight)
shape_cut = '({c}) * {we}'.format(c=shape_cut, we=weight)
ttree.Project(hname, vcfg.drawname, norm_cut)
if shape_cut != norm_cut:
scale = hist.Integral()
ttree.Project(hname, vcfg.drawname, shape_cut)
hist.Scale(scale/hist.Integral())
stack = all_stack or (not cfg.is_data and not cfg.is_signal)
hist.Scale(cfg.scale)
if cfg.name in plot:
# print 'Histogram', cfg.name, 'already exists; adding...', cfg.dir_name
hist_to_add = Histogram(cfg.name, hist)
if (not cfg.is_data) and (not cfg.is_dde):
hist_to_add.SetWeight(hist_cfg.lumi*cfg.xsec/cfg.sumweights)
if cfg.is_dde:
# hist_to_add.SetWeight(hist_to_add.obj.GetEntries()/cfg.sumweights)
hist_to_add.SetWeight(cfg.sumweights)
plot[cfg.name].Add(hist_to_add)
else:
plot_hist = plot.AddHistogram(cfg.name, hist, stack=stack)
if (not cfg.is_data) and (not cfg.is_dde):
plot_hist.SetWeight(hist_cfg.lumi*cfg.xsec/cfg.sumweights)
if cfg.is_dde:
# plot_hist.SetWeight(plot_hist.obj.GetEntries()/cfg.sumweights)
plot_hist.SetWeight(cfg.sumweights)
plot._ApplyPrefs()
return plot
def createHistograms(self, hist_cfg, all_stack=False, verbose=False, friend_func=None, vcfgs=None, multiprocess = True):
'''Method to create actual histogram (DataMCPlot) instances from histogram
config; this version handles multiple variables via MultiDraw.
'''
print('###########################################################')
print('# creating histograms for %i sample(s)...'%len(self.hist_cfg.cfgs))
if multiprocess == True:
print('# multiprocess-mode: ON')
if multiprocess == False:
print('# multiprocess-mode: OFF')
print('###########################################################')
if multiprocess == True:
#using multiprocess to create the histograms
pool = Pool(processes=len(self.hist_cfg.cfgs))
result = pool.map(self.makealltheplots, self.hist_cfg.cfgs)
# if we don't use multiprocess, we compute the histos one by one - good for debugging.
if multiprocess == False:
for i, cfg in enumerate(self.hist_cfg.cfgs):
# result = self.makealltheplots(self.hist_cfg.cfgs[i])
try:
result = self.makealltheplots(self.hist_cfg.cfgs[i])
except:
set_trace()
# workers = cpu_count()
# result = []
# batches = len(self.hist_cfg.cfgs) / workers + 1 if len(self.hist_cfg.cfgs) % workers != 0 else len(self.hist_cfg.cfgs) / workers
# pool = Pool(processes=batches)
# print('number of batches for filling histos (%i samples each): %i'%(workers, batches))
# for i in range(batches):
# try: histlist = self.hist_cfg.cfgs[i*workers:(i+1)*workers]
# except: print('entering exception'); histlist = self.hist_cfg.cfgs[(batches-1)*workers:len(self.hist_cfg.cfgs)-1]
# result += pool.map(self.makealltheplots, histlist)
for i, cfg in enumerate(self.hist_cfg.cfgs):
stack = not cfg.is_data and not cfg.is_signal
# print(cfg.name, stack)
for vcfg in self.vcfgs:
try:
hist = result[i][vcfg.name].histos[0].obj # result[0]['CR_hnl_m_12'].histos[0]
except:
try:
hist = result[vcfg.name].histos[0].obj # result[0]['CR_hnl_m_12'].histos[0]
except:
set_trace()
# hist = hists[vcfg.name]
plot = self.plots[vcfg.name]
# hist.Scale(cfg.scale)
if cfg.name in plot:
# print 'Histogram', cfg.name, 'already exists; adding...', cfg.dir_name
hist_to_add = Histogram(cfg.name, hist)
if (not cfg.is_data) and (not cfg.is_dde):
hist_to_add.SetWeight(hist_cfg.lumi*cfg.xsec/cfg.sumweights)
plot[cfg.name].Add(hist_to_add)
if cfg.is_dde:
try:
# hist_to_add.SetWeight(hist_to_add.obj.GetEntries()/cfg.sumweights)
hist_to_add.SetWeight(cfg.sumweights)
except:
set_trace()
else:
# print(cfg.name, hist.GetEntries(), stack)
plot_hist = plot.AddHistogram(cfg.name, hist, stack=stack)
# print('added histo %s'%vcfg.name)
if (not cfg.is_data) and (not cfg.is_dde):
plot_hist.SetWeight(self.hist_cfg.lumi*cfg.xsec/cfg.sumweights)
if cfg.is_dde:
# plot_hist.SetWeight(plot_hist.obj.GetEntries()/cfg.sumweights)
plot_hist.SetWeight(cfg.sumweights)
# print(cfg.name, vcfg.name, len(plot.histos))
print('###########################################################')
print('# initializing histos done, making stacks...')
for plot in self.plots.itervalues():
try:
plot._ApplyPrefs()
except:
set_trace() #if this error is raised, check in HNLStyle.py whether the style is defined coffectly
print('# number of plots to draw: %i'%len(self.plots))
print('###########################################################')
procs = []
for i, plot in enumerate(self.plots.itervalues()):
proc = Process(target=plot.Draw, args=())
procs.append(proc)
proc.start()
for proc in procs:
proc.join()
# for plot in self.plots.itervalues():
# plot._ApplyPrefs()
# plot.Draw()
return self.plots
def createHistograms(hist_cfg,
all_stack=False,
verbose=False,
friend_func=None):
'''Method to create actual histogram (DataMCPlot) instances from histogram
config; this version handles multiple variables via MultiDraw.
'''
vcfgs = hist_cfg.vars
plots = {}
for vcfg in vcfgs:
plot = DataMCPlot(vcfg.name)
plot.lumi = hist_cfg.lumi
if vcfg.name in plots:
print 'Adding variable with same name twice', vcfg.name, 'not yet foreseen; taking the last'
plots[vcfg.name] = plot
for cfg in hist_cfg.cfgs:
# First check whether it's a sub-histo or not
if isinstance(cfg, HistogramCfg):
hists = createHistograms(cfg, all_stack=True)
for vcfg in vcfgs:
hist = hists[vcfg.name]
plot = plots[vcfg.name]
hist._BuildStack(hist._SortedHistograms(), ytitle='Events')
total_hist = plot.AddHistogram(cfg.name,
hist.stack.totalHist.weighted,
stack=True)
if cfg.norm_cfg is not None:
norm_hist = createHistogram(cfg.norm_cfg, all_stack=True)
norm_hist._BuildStack(norm_hist._SortedHistograms(),
ytitle='Events')
total_hist.Scale(hist.stack.integral / total_hist.Yield())
if cfg.total_scale is not None:
total_hist.Scale(cfg.total_scale)
# print 'Scaling total', hist_cfg.name, 'by', cfg.total_scale
else:
# It's a sample cfg
# Now read the tree
file_name = '/'.join(
[cfg.ana_dir, cfg.dir_name, cfg.tree_prod_name, 'tree.root'])
# attach the trees to the first DataMCPlot
plot = plots[vcfgs[0].name]
ttree = plot.readTree(file_name,
cfg.tree_name,
verbose=verbose,
friend_func=friend_func)
norm_cut = hist_cfg.cut
shape_cut = hist_cfg.cut
if cfg.norm_cut:
norm_cut = cfg.norm_cut
if cfg.shape_cut:
shape_cut = cfg.shape_cut
weight = hist_cfg.weight
if cfg.weight_expr:
weight = '*'.join([weight, cfg.weight_expr])
if hist_cfg.weight:
norm_cut = '({c}) * {we}'.format(c=norm_cut, we=weight)
shape_cut = '({c}) * {we}'.format(c=shape_cut, we=weight)
# Initialise all hists before the multidraw
hists = {}
for vcfg in vcfgs:
# plot = plots[vcfg.name]
hname = '_'.join(
[hist_cfg.name, cfg.name, vcfg.name, cfg.dir_name])
if 'xmin' in vcfg.binning:
hist = TH1F(hname, '', vcfg.binning['nbinsx'],
vcfg.binning['xmin'], vcfg.binning['xmax'])
else:
hist = TH1F(hname, '', len(vcfg.binning) - 1, vcfg.binning)
initHist(hist, vcfg)
hists[vcfg.name] = hist
var_hist_tuples = []
for vcfg in vcfgs:
# var_hist_tuples.append(('{var} >> {hist}'.format(var=vcfg.drawname, hist=hists[vcfg.name].GetName()), '1.'))
var_hist_tuples.append('{var} >> {hist}'.format(
var=vcfg.drawname, hist=hists[vcfg.name].GetName()))
# Implement the multidraw.
ttree.MultiDraw(var_hist_tuples, norm_cut)
# Do another multidraw here, if needed, and reset the scales in a separate loop
if shape_cut != norm_cut:
scale = hist.Integral()
ttree.Project(hname, vcfg.drawname, shape_cut)
hist.Scale(scale / hist.Integral())
stack = all_stack or (not cfg.is_data and not cfg.is_signal)
# Loop again over the variables and add histograms to plots one by one
for vcfg in vcfgs:
hist = hists[vcfg.name]
plot = plots[vcfg.name]
hist.Scale(cfg.scale)
if cfg.name in plot:
print 'Histogram', cfg.name, 'already exists; adding...'
plot[cfg.name].Add(Histogram(cfg.name, hist))
else:
plot_hist = plot.AddHistogram(cfg.name, hist, stack=stack)
if not cfg.is_data:
plot_hist.SetWeight(hist_cfg.lumi * cfg.xsec /
cfg.sumweights)
for plot in plots.itervalues():
plot._ApplyPrefs()
return plots
def makealltheplots(self, cfg):
verbose=False
friend_func=None
all_stack=False
# for cfg in [hist_cfg.cfgs[0]]:
# for cfg in hist_cfg.cfgs:
# First check whether it's a sub-histo or not
if isinstance(cfg, HistogramCfg):
hists = createHistograms(cfg, all_stack=True, vcfgs=self.vcfgs)
for h in hists: print(h)
for vcfg in self.vcfgs:
hist = hists[vcfg.name]
plot = self.plots[vcfg.name]
set_trace()
hist._BuildStack(hist._SortedHistograms(), ytitle='Events')
print('stack built')
total_hist = plot.AddHistogram(cfg.name, hist.stack.totalHist.weighted, stack=True)
if cfg.norm_cfg is not None:
norm_hist = createHistogram(cfg.norm_cfg, all_stack=True)
norm_hist._BuildStack(norm_hist._SortedHistograms(), ytitle='Events')
total_hist.Scale(hist.stack.integral/total_hist.Yield())
if cfg.total_scale is not None:
total_hist.Scale(cfg.total_scale)
# print 'Scaling total', hist_cfg.name, 'by', cfg.total_scale
else:
print('building histos for %s'%cfg.name)
# It's a sample cfg
# Now read the tree
file_name = '/'.join([cfg.ana_dir, cfg.dir_name, cfg.tree_prod_name, 'tree.root'])
# attach the trees to the first DataMCPlot
plot = self.plots[self.vcfgs[0].name]
# set_trace()
ttree = plot.readTree(file_name, cfg.tree_name, verbose=verbose, friend_func=friend_func)
norm_cut = self.hist_cfg.cut
shape_cut = self.hist_cfg.cut
if cfg.norm_cut:
norm_cut = cfg.norm_cut
if cfg.shape_cut:
shape_cut = cfg.shape_cut
weight = self.hist_cfg.weight
if cfg.weight_expr:
weight = '*'.join([weight, cfg.weight_expr])
if cfg.cut_replace_func:
norm_cut = cfg.cut_replace_func(norm_cut)
shape_cut = cfg.cut_replace_func(norm_cut)
if self.hist_cfg.weight:
norm_cut = '({c}) * {we}'.format(c=norm_cut, we=weight)
shape_cut = '({c}) * {we}'.format(c=shape_cut, we=weight)
# print '#### FULL CUT ####', norm_cut
# Initialise all hists before the multidraw
hists = {}
for vcfg in self.vcfgs:
# plot = self.plots[vcfg.name]
hname = '_'.join([self.hist_cfg.name, hashlib.md5(self.hist_cfg.cut).hexdigest(), cfg.name, vcfg.name, cfg.dir_name])
if any(str(b) == 'xmin' for b in vcfg.binning):
hist = TH1F(hname, '', vcfg.binning['nbinsx'],
vcfg.binning['xmin'], vcfg.binning['xmax'])
else:
hist = TH1F(hname, '', len(vcfg.binning)-1, vcfg.binning)
initHist(hist, vcfg)
hists[vcfg.name] = hist
var_hist_tuples = []
for vcfg in self.vcfgs:
# var_hist_tuples.append(('{var} >> {hist}'.format(var=vcfg.drawname, hist=hists[vcfg.name].GetName()), '1.'))
# pool.map(var_hist_tuples.append,('{var} >> {hist}'.format(var=vcfg.drawname, hist=hists[vcfg.name].GetName())))
var_hist_tuples.append('{var} >> {hist}'.format(var=vcfg.drawname, hist=hists[vcfg.name].GetName()))
# Implement the multidraw.
ttree.MultiDraw(var_hist_tuples, norm_cut)
# Do another multidraw here, if needed, and reset the scales in a separate loop
if shape_cut != norm_cut:
scale = hist.Integral()
ttree.Project(hname, vcfg.drawname, shape_cut)
hist.Scale(scale/hist.Integral())
stack = all_stack or (not cfg.is_data and not cfg.is_signal)
# Loop again over the variables and add histograms to self.plots one by one
for vcfg in self.vcfgs:
hist = hists[vcfg.name]
plot = self.plots[vcfg.name]
hist.Scale(cfg.scale)
if cfg.name in plot:
print 'Histogram', cfg.name, 'already exists; adding...', cfg.dir_name
hist_to_add = Histogram(cfg.name, hist)
if not cfg.is_data:
hist_to_add.SetWeight(hist_cfg.lumi*cfg.xsec/cfg.sumweights)
plot[cfg.name].Add(hist_to_add)
else:
# print(cfg.name, hist.GetEntries(), stack)
plot_hist = plot.AddHistogram(cfg.name, hist, stack=stack)
# print('added histo %s for %s'%(vcfg.name,cfg.name))
if not cfg.is_data:
plot_hist.SetWeight(self.hist_cfg.lumi*cfg.xsec/cfg.sumweights)
print('added histos for %s'%cfg.name)
PLOTS = self.plots
return PLOTS
def createHistogram(hist_cfg, all_stack=False, verbose=False):
'''Method to create actual histogram (DataMCPlot) instance from histogram
config.
'''
plot = DataMCPlot(hist_cfg.var.name)
plot.lumi = hist_cfg.lumi
vcfg = hist_cfg.var
for cfg in hist_cfg.cfgs:
# First check whether it's a sub-histo or not
if isinstance(cfg, HistogramCfg):
hist = createHistogram(cfg, all_stack=True)
hist._BuildStack(hist._SortedHistograms(), ytitle='Events')
total_hist = plot.AddHistogram(cfg.name,
hist.stack.totalHist.weighted,
stack=True)
if cfg.norm_cfg is not None:
norm_hist = createHistogram(cfg.norm_cfg, all_stack=True)
norm_hist._BuildStack(norm_hist._SortedHistograms(),
ytitle='Events')
total_hist.Scale(hist.stack.integral / total_hist.Yield())
if cfg.total_scale is not None:
total_hist.Scale(cfg.total_scale)
else:
# It's a sample cfg
hname = '_'.join(
[hist_cfg.name, cfg.name, vcfg.name, cfg.dir_name])
if 'xmin' in vcfg.binning:
hist = TH1F(hname, '', vcfg.binning['nbinsx'],
vcfg.binning['xmin'], vcfg.binning['xmax'])
else:
hist = TH1F(hname, '', len(vcfg.binning) - 1, vcfg.binning)
initHist(hist, vcfg)
file_name = '/'.join(
[cfg.ana_dir, cfg.dir_name, cfg.tree_prod_name, 'tree.root'])
ttree = plot.readTree(file_name, cfg.tree_name, verbose=verbose)
norm_cut = hist_cfg.cut
shape_cut = hist_cfg.cut
if cfg.norm_cut:
norm_cut = cfg.norm_cut
if cfg.shape_cut:
shape_cut = cfg.shape_cut
weight = hist_cfg.weight
if cfg.weight_expr:
weight = '*'.join([weight, cfg.weight_expr])
if hist_cfg.weight:
norm_cut = '({c}) * {we}'.format(c=norm_cut, we=weight)
shape_cut = '({c}) * {we}'.format(c=shape_cut, we=weight)
ttree.Project(hname, vcfg.drawname, norm_cut)
if shape_cut != norm_cut:
scale = hist.Integral()
ttree.Project(hname, vcfg.drawname, shape_cut)
hist.Scale(scale / hist.Integral())
stack = all_stack or (not cfg.is_data and not cfg.is_signal)
hist.Scale(cfg.scale)
if cfg.name in plot:
plot[cfg.name].Add(Histogram(cfg.name, hist))
else:
plot_hist = plot.AddHistogram(cfg.name, hist, stack=stack)
if not cfg.is_data:
plot_hist.SetWeight(hist_cfg.lumi * cfg.xsec / cfg.sumweights)
plot._ApplyPrefs()
return plot
